Microwave-based breast imaging is a developing technique that has demonstrated potential as a method for breast cancer detection. The modality interrogates the breast tissues using low-power non-ionizing microwave radiation and relies on a contrast in the dielectric properties of malignant and healthy breast tissues. Research into this field has been motivated by the low cost and small size of microwave hardware, the large contrast in the dielectric properties of malignant and healthy tissues, and the favourable safety profile of non-ionizing microwave imaging. The modality has the potential to be used for screening in remote and under-serviced communities due to these factors. Two major challenges limit breast microwave imaging. Existing estimates of the diagnostic performance, particularly the diagnostic specificity, are relatively poor due to a lack of accurate image reconstruction algorithms. Additionally, image quality analysis has been limited to evaluations of image contrast and localization error that rely on single-pixel intensities. The lack of informative measures of image quality has limited the development of robust image reconstruction methods in the field of research. This thesis addresses these two challenges through the development of a novel image reconstruction method capable of enhanced physics modelling and through the development of methods, metrics, and specialized phantoms for image quality analysis. Enhanced physics modelling improves the physics model used to reconstruct an image and was observed to improve image accuracy and diagnostic accuracy. The standardized methods, metrics, and phantoms developed for image quality analysis were used to compare various breast microwave imaging systems and image reconstruction methods.